Patent Eligibility Of Neural MicrochIP Architecture For SustAInable AI Computation.

20 Mar 2026 --
0 Comments

1. Patent Eligibility Framework in the U.S.

Under 35 U.S.C. § 101, patentable subject matter includes:

"any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof."

However, the Supreme Court has carved out exceptions for:

Abstract ideas
Laws of nature
Natural phenomena

Neural microchip architectures, especially for AI, often face scrutiny under the “abstract idea” exception because software, algorithms, and AI models can be considered abstract unless tied to a specific machine or technological improvement.

2. Key Case Laws on Patent Eligibility

Here are five pivotal cases that directly impact patent eligibility for AI hardware/software innovations:

Case 1: Alice Corp. v. CLS Bank International (2014)

Citation: 573 U.S. 208 (2014)

Facts:
Alice Corp. claimed patents for a computerized system for mitigating settlement risk in financial transactions. The question was whether their method implemented on a generic computer was patentable.

Holding:

The Supreme Court ruled that implementing an abstract idea on a computer is not enough.
Two-step framework (Alice/Mayo test):
1. Determine if the claim is directed to a patent-ineligible concept (abstract idea, law of nature, or natural phenomenon).
2. If yes, determine whether the claim contains an “inventive concept” that transforms it into patent-eligible subject matter.

Application to Neural Microchips:

A neural microchip for sustainable AI that merely executes an AI algorithm might be considered an abstract idea if it just runs a standard neural network.
To be patentable, the microchip must show a specific technological improvement, like a novel architecture that reduces power consumption, improves speed, or reduces heat dissipation.

Case 2: Diamond v. Diehr (1981)

Citation: 450 U.S. 175 (1981)

Facts:
Diehr patented a process for curing rubber using a mathematical formula and a computer to control a press.

Holding:

The Court held that a process that applies a mathematical formula in a practical application is patentable.
The key was that the patent improved a technological process, not the formula itself.

Application to Neural Microchips:

If a microchip architecture optimizes AI computations in a way that improves energy efficiency or hardware performance, similar to how Diehr’s process improved rubber curing, it can be considered patent-eligible.
Claiming the architecture as a whole system for sustainable AI, rather than the algorithm alone, strengthens eligibility.

Case 3: Enfish, LLC v. Microsoft Corp. (2016)

Citation: 822 F.3d 1327 (Fed. Cir. 2016)

Facts:
Enfish claimed a self-referential table for databases. Microsoft argued it was an abstract idea.

Holding:

The Federal Circuit held that claims directed to a specific improvement in computer functionality can be patent-eligible.
Focus on whether the invention improves the functioning of the computer itself.

Application to Neural Microchips:

A neural microchip with architecture designed for sustainable AI computation—reducing energy consumption, increasing speed, or enabling more parallel operations—could be considered a technological improvement, satisfying Enfish criteria.

Case 4: McRO, Inc. v. Bandai Namco Games America Inc. (2016)

Citation: 837 F.3d 1299 (Fed. Cir. 2016)

Facts:
McRO patented automated lip-sync animation techniques using rules applied to phonemes.

Holding:

Patent claims not abstract because they automated a previously manual process in a specific way using rules.
Court emphasized that implementation details matter.

Application to Neural Microchips:

If a neural microchip implements a novel computation or optimization method to achieve sustainability (e.g., low-power AI inference), the architecture could be patent-eligible.
Mere generic implementation of AI algorithms is not enough; specific hardware features or methods are critical.

Case 5: Ariosa Diagnostics, Inc. v. Sequenom, Inc. (2015)

Citation: 788 F.3d 1371 (Fed. Cir. 2015)

Facts:
Sequenom patented methods for detecting fetal DNA in maternal blood.

Holding:

Court ruled claims were directed to a natural phenomenon (cell-free fetal DNA) and lacked inventive concept.
Illustrates that discovery alone is not patentable; practical application must exist.

Application to Neural Microchips:

Simply observing that neural networks can run on low-power chips is not patentable.
The patent must claim how the architecture achieves sustainability in AI computation (e.g., design improvements, novel circuits).

3. Key Takeaways for Neural Microchip Patents

To ensure patent eligibility for sustainable AI microchips, the claims should:

Emphasize hardware innovation:
- Novel transistor layout, power gating techniques, low-energy matrix multipliers, etc.
Link software and hardware improvement:
- Example: AI inference optimized for edge devices with minimal energy.
Demonstrate technological improvement:
- Not just using AI on a chip, but achieving measurable improvement in performance, power, or thermal efficiency.
Draft claims carefully:
- Avoid claims limited to abstract algorithms.
- Include methods, systems, and hardware elements.

Summary Table: Cases and Lessons

Case	Key Principle	Lesson for Neural Microchips
Alice v. CLS Bank	Abstract idea test (2-step)	Show inventive concept beyond AI algorithm
Diehr	Practical application of formula is patentable	Highlight technological improvements in chip design
Enfish	Specific computer improvement = patentable	Emphasize improved AI computation efficiency
McRO	Implementation details matter	Novel automation in hardware execution strengthens eligibility
Ariosa	Natural phenomena not patentable	Observation of AI efficiency alone insufficient

Conclusion

A neural microchip architecture for sustainable AI computation can be patentable if it:

Represents a specific, tangible technological improvement.
Goes beyond abstract AI algorithms to concrete hardware innovation.
Demonstrates measurable benefits in power consumption, speed, or reliability.

This aligns with the precedents of Alice, Diehr, Enfish, McRO, and Ariosa, forming a solid foundation for patent strategy.